AmazonS3XMLParser.php

<?php

require_once 'EasyXMLParser.php';

class AmazonS3XMLParser
{
	protected $properties;
	protected $contents;
	protected $currentItem;

	function __construct($filename) {
		$this->contents = array();

		$root = 'listbucketresult';

		$parser = new EasyXMLParser;

		$parser->rule("$root ?", array(
			'cdata' => function ($data, $parser) {
				$stack = $parser->stack();
				$tag = end($stack);
				@$this->properties[$tag['name']] .= $data;
			}
		))
		->rule("$root contents", array(
			'start' => function () {
				$this->currentItem = array();
			},
			'end' => function () {
				$this->contents[] = $this->currentItem;
				$this->currentItem = null;
				unset ($this->currentItem);
			}
		))
		->rule("$root contents ?", array(
			'cdata' => function ($data, $parser) {
				$stack = $parser->stack();
				$tag = end($stack);
				@$this->currentItem[$tag['name']] .= $data;
			}
		));

		$parser->parse_file($filename);
	}

	function __get($name) {
		$name = strtolower($name);

		if ($name == 'contents') {
			return $this->contents;
		}

		if (array_key_exists($name, $this->properties)) {
			return $this->properties[$name];
		}
	}
}

EasyXMLParser.php

<?php

require_once 'EasyXMLParserException.php';
require_once 'Finalizer.php';

class EasyXMLParser
{
	protected $stack;              # stack of tags and attributes
	protected $hstack;             # stack of tag handlers
	protected $wstack;             # '+' handlers
	protected $ruleset = array();  # ruleset
	protected $miss;               # current depth of missed tags

	public $bufferSize = 4096;

	public function parse_file($filename)
	{
		$this->stack = array();
		$this->hstack = array($this->ruleset);
		$this->wstack = array();
		$this->miss = 0;

		$fin = new Finalizer('set_error_handler', set_error_handler([$this, 'error_handler']));

		$fh = fopen($filename, 'r');

		if (!$fh) {
			return false;
		}

		$fin('fclose', $fh);

		$parser = xml_parser_create();

		$fin('xml_parser_free', $parser);

		xml_set_object($parser, $this);
		xml_set_character_data_handler($parser, 'character_data_handler');
		xml_set_default_handler($parser, 'default_handler');
		xml_set_element_handler($parser, 'start_element_handler', 'end_element_handler');

		do {
			try {
				$buffer = fread($fh, $this->bufferSize);
				$last = feof($fh)
				xml_parse($parser, $buffer, $last);
			} catch (Exception $e) {
				$line = xml_get_current_line_number($parser);
				$column = xml_get_current_column_number($parser);
				$message = "Parse error on line $line, column $column";
				$code = $e->code();
				throw new EasyXMLParserException($message, $code, $line, $column, $e);
			}
		} while (!$last);

		return true;
	}

	public function rule($tagset, $rules)
	{
		foreach ((array) $tagset as $tags) {
			$tags = explode(' ', $tags);

			$handler = &$this->ruleset;

			foreach ($tags as $tag) {
				$tag = strtolower($tag);
				if ($tag !== '') {
					if (!array_key_exists($tag, $handler) || !is_array($handler[$tag])) {
						$handler[$tag] = array();
					}
					$handler = &$handler[$tag];
				}
			}

			foreach ($rules as $key => $callback) {
				$handler["@$key"] = $callback;
			}
		}
	}

	public function stack()
	{
		return $this->stack;
	}

	public function path()
	{
		return implode('/', array_map(function($tag) {
			return $tag['name'];
		}, $this->stack());
	}

	protected function character_data_handler($parser, $data)
	{
		if (!$this->miss) {
			$handler = end($this->hstack);

			if (array_key_exists('@cdata', $handler) && is_callable($handler['@cdata'])) {
				$handler['@cdata']($data, $this);
			}
		}
	}

	protected function default_handler($parser, $data)
	{
		if (!$this->miss) {
			$handler = end($this->hstack);

			if ($handler && array_key_exists('@default', $handler) && is_callable($handler['@default'])) {
				$handler['@default']($data, $this);
			}
		}
	}

	protected function start_element_handler($parser, $name, $attr)
	{
		if ($this->miss) {
			$this->miss++;
		} else {
			$attr = array_change_key_case($attr);
			$name = strtolower($name);

			$handler = end($this->hstack);

			if (array_key_exists('+', $handler) && is_array($handler['+'])) {
				$this->wstack[] = $handler['+'];
			}

			if (array_key_exists($name, $handler) && is_array($handler[$name])) {
				$handler = $handler[$name];
			} elseif (array_key_exists('?', $handler) && is_array($handler['?'])) {
				$handler = $handler['?'];
			} elseif ($this->wstack) {
				$handler = end($this->wstack);
			} else {
				$this->miss++;
			}

			if (!$this->miss) {
				$this->stack[] = array('name' => $name, 'attr' => $attr);
				$this->hstack[] = $handler;

				if (array_key_exists('@start', $handler) && is_callable($handler['@start'])) {
					$handler['@start']($attr, $name, $this);
				}
			}
		}
	}

	protected function end_element_handler($parser, $name)
	{
		if ($this->miss) {
			$this->miss--;
		} else {
			$handler = array_pop($this->hstack);
			$tag = array_pop($this->stack);

			if (array_key_exists('@end', $handler) && is_callable($handler['@end'])) {
				$handler['@end']($tag['attr'], $tag['name'], $this);
			}

			$handler = end($this->hstack);
			if (array_key_exists('+', $handler) && is_array($handler['+'])) {
				array_pop($this->wstack);
			}
		}
	}

	protected function error_handler($errno, $errstr, $errfile, $errline)
	{
		throw new ErrorException($errstr, $errno, E_ERROR, $errfile, $errline);
	}
}

EasyXMLParserException.php

<?php

class EasyXMLParserException extends Exception
{
	protected $xmlLine;
	protected $xmlColumn;

	public function __construct($message, $code, $line, $column, Throwable $previous)
	{
		parent::__construct($message, $code, $previous);
		$this->xmlLine = $line;
		$this->xmlColumn = $column;
	}
	
	public function getXMLLine()
	{
		return $this->xmlLine;
	}
	
	public function getXMLColumn;
	{
		return $this->xmlColumn;
	}
}

Finalizer.php

<?php

class Finalizer
{
	protected $callbacks = array();

	public function __construct(callable $callback = null)
	{
		if ($callback) {
			$this->callbacks[] = func_get_args();
		}
	}

	public function __destruct()
	{
		foreach ($this->callbacks as $callback) {
			call_user_func_array('call_user_func', $callback);
		}
	}

	public function __invoke(callable $callback)
	{
		$this->callbacks[] = func_get_args();
	}
}

NextoXMLParser.php

<?php

require_once 'EasyXMLParser.inc';

class NextoXMLParser
{
	protected $parser;

	protected $product;
	protected $tag;
	protected $obj;
	protected $subscription;
	protected $value;

	public $product_handler;

	public function __construct()
	{
		$this->parser = new EasyXMLParser;

		$handlers['@start'] = array($this, 'product_start');
		$handlers['@end'] = array($this, 'product_end');
		$handlers['*']['@start'] = array($this, 'unknown_start');
		$handlers['*']['@end'] = array($this, 'unknown_end');

		$tags = array(
			'name',
			'type',
			'url',
			'category',
			'issn',
			'manufacturer',
			'language',
			'description',
			'review-note',
		);

		foreach ($tags as $tag)
		{
			$handlers[$tag]['@start']      = array($this, 'product_field_start');
			$handlers[$tag]['@end']        = array($this, 'product_field_end');
			$handlers[$tag]['@cdata']      = array($this, 'product_field_cdata');
			$handlers[$tag]['*']['@start'] = array($this, 'unknown_start');
			$handlers[$tag]['*']['@end']   = array($this, 'unknown_end');
		}

		$handlers['issues']['@start']             = array($this, 'issues_start');
		$handlers['issues']['@end']               = array($this, 'issues_end');
		$handlers['issues']['*']['@start']        = array($this, 'unknown_start');
		$handlers['issues']['*']['@end']          = array($this, 'unknown_end');

		$handlers['subscriptions']['@start']      = array($this, 'subscriptions_start');
		$handlers['subscriptions']['@end']        = array($this, 'subscriptions_end');
		$handlers['subscriptions']['*']['@start'] = array($this, 'unknown_start');
		$handlers['subscriptions']['*']['@end']   = array($this, 'unknown_end');

		$issue['@start']      = array($this, 'issue_start');
		$issue['@end']        = array($this, 'issue_end');
		$issue['*']['@start'] = array($this, 'unknown_start');
		$issue['*']['@end']   = array($this, 'unknown_end');

		$tags = array(
			'title',
			'publication-date',
			'activation-date',
			'formats-all',
			'description',
			'pages-count',
			'audiobook-length',
		);

		foreach ($tags as $tag)
		{
			$issue[$tag]['@start']      = array($this, 'issue_field_start');
			$issue[$tag]['@end']        = array($this, 'issue_field_end');
			$issue[$tag]['@cdata']      = array($this, 'issue_field_cdata');
			$issue[$tag]['*']['@start'] = array($this, 'unknown_start');
			$issue[$tag]['*']['@end']   = array($this, 'unknown_end');
		}

		$tags = array(
			'price-original',
			'price-normal',
			'price-club',
		);

		foreach ($tags as $tag) {
			$issue['price'][$tag]['@start']      = array($this, 'issue_field_start');
			$issue['price'][$tag]['@end']        = array($this, 'issue_field_end');
			$issue['price'][$tag]['@cdata']      = array($this, 'issue_field_cdata');
			$issue['price'][$tag]['*']['@start'] = array($this, 'unknown_start');
			$issue['price'][$tag]['*']['@end']   = array($this, 'unknown_end');
		}

		$issue['price']['*']['@start'] = array($this, 'unknown_start');
		$issue['price']['*']['@end']   = array($this, 'unknown_end');

		$tags = array(
			'formats'        => 'format',
			'covers'         => 'cover',
			'tocs'           => 'table-of-content',
			'authors'        => 'author',
			'lectors'        => 'lector',
			'free-fragments' => 'free-fragment',
		);

		foreach ($tags as $k => $v) {
			$issue[$k]['@start']          = array($this, 'array_start');
			$issue[$k]['@end']            = array($this, 'array_end');
			$issue[$k]['*']['@start']     = array($this, 'unknown_start');
			$issue[$k]['*']['@end']       = array($this, 'unknown_end');
			$issue[$k][$v]['@start']      = array($this, 'value_start');
			$issue[$k][$v]['@end']        = array($this, 'value_end');
			$issue[$k][$v]['@cdata']      = array($this, 'value_cdata');
			$issue[$k][$v]['*']['@start'] = array($this, 'unknown_start');
			$issue[$k][$v]['*']['@end']   = array($this, 'unknown_end');
		}

		$subscription['@start']      = array($this, 'subscription_start');
		$subscription['@end']        = array($this, 'subscription_end');
		$subscription['*']['@start'] = array($this, 'unknown_start');
		$subscription['*']['@end']   = array($this, 'unknown_end');

		$tags = array(
			'name',
			'quantity',
			'format',
			'cyclic',
		);

		foreach ($tags as $tag) {
			$subscription[$tag]['@start']      = array($this, 'subscription_field_start');
			$subscription[$tag]['@end']        = array($this, 'subscription_field_end');
			$subscription[$tag]['@cdata']      = array($this, 'subscription_field_cdata');
			$subscription[$tag]['*']['@start'] = array($this, 'unknown_start');
			$subscription[$tag]['*']['@end']   = array($this, 'unknown_end');
		}

		$tags = array(
			'price-original',
			'price-normal',
			'price-club',
		);

		foreach ($tags as $tag) {
			$subscription['price'][$tag]['@start']      = array($this, 'subscription_field_start');
			$subscription['price'][$tag]['@end']        = array($this, 'subscription_field_end');
			$subscription['price'][$tag]['@cdata']      = array($this, 'subscription_field_cdata');
			$subscription['price'][$tag]['*']['@start'] = array($this, 'unknown_start');
			$subscription['price'][$tag]['*']['@end']   = array($this, 'unknown_end');
		}

		$subscription['price']['*']['@start'] = array($this, 'unknown_start');
		$subscription['price']['*']['@end']   = array($this, 'unknown_end');

		$handlers['issues']['issue'] = $issue;
		$handlers['subscriptions']['subscription'] = $subscription;

		$this->parser->handlers['nexto-products']['product']     = $handlers;
		$this->parser->handlers['nexto-products']['*']['@start'] = array($this, 'unknown_start');
		$this->parser->handlers['nexto-products']['*']['@end']   = array($this, 'unknown_start');
		$this->parser->handlers['*']['@start']                   = array($this, 'unknown_start');
		$this->parser->handlers['*']['@end']                     = array($this, 'unknown_end');
	}

	public function parse_file($filename)
	{
		return $this->parser->parse_file($filename);
	}

	public function product_start($parser, $name, $attr)
	{
		$this->product = array('attr' => $attr, 'data' => array());
	}

	public function product_end($parser, $name)
	{
		call_user_func($this->product_handler, $this->product);
		unset($this->product);
	}

	public function product_field_start($parser, $name, $attr)
	{
		if (isset($this->product['data'][$name])) {
			throw new Exception("Duplicate $name field in <product>");
		}

		$this->tag = $name;
		$this->product['data'][$name] = array('attr' => $attr, 'value' => '');
	}

	public function product_field_end($parser, $name)
	{
		unset($this->tag);
	}

	public function product_field_cdata($parser, $data)
	{
		$this->product['data'][$this->tag]['value'] .= $data;
	}

	public function issues_start($parser, $name, $attr)
	{
		if (isset($this->product['data']['issues'])) {
			throw new Exception('<issues> has been already defined');
		}

		$this->product['data']['issues'] = array('attr' => $attr, 'list' => array());
	}

	public function issues_end($parser, $name)
	{
		# nothing
	}

	public function issue_start($parser, $name, $attr)
	{
		$this->obj = array();
	}

	public function issue_end($parser, $name)
	{
		$this->product['data']['issues']['list'][] = $this->obj;
		unset($this->obj);
	}

	public function issue_field_start($parser, $name, $attr)
	{
		if (isset($this->obj[$name])) {
			throw new Exception("Duplicate $name field in <issue>");
		}

		$this->tag = $name;
		$this->obj[$name] = array('attr' => $attr, 'value' => '');
	}

	public function issue_field_end($parser, $name)
	{
		unset($this->tag);
	}

	public function issue_field_cdata($parser, $data)
	{
		$this->obj[$this->tag]['value'] .= $data;
	}

	public function subscriptions_start($parser, $name, $attr)
	{
		if (isset($this->product['data']['subscriptions'])) {
			throw new Exception('<subscriptions> has been already defined');
		}

		$this->product['data']['subscriptions'] = array('attr' => $attr, 'list' => array());
	}

	public function subscriptions_end($parser, $name)
	{
		# nothing
	}

	public function subscription_start($parser, $name, $attr)
	{
		$this->obj = array();
	}

	public function subscription_end($parser, $name)
	{
		$this->product['data']['subscriptions']['list'][] = $this->obj;
		unset($this->obj);
	}

	public function subscription_field_start($parser, $name, $attr)
	{
		if (isset($this->obj[$name])) {
			throw new Exception("Duplicate $name field in <subscription>");
		}

		$this->tag = $name;
		$this->obj[$name] = array('attr' => $attr, 'value' => '');
	}

	public function subscription_field_end($parser, $name)
	{
		unset ($this->tag);
	}

	public function subscription_field_cdata($parser, $data)
	{
		$this->obj[$this->tag]['value'] .= $data;
	}

	public function array_start($parser, $name, $attr)
	{
		if (isset($this->obj[$name]))
			throw new Exception("<$name> has been already defined");

		$this->obj[$name] = array('attr' => $attr, 'list' => array());

		$this->tag = $name;
	}

	public function array_end($parser, $name)
	{
		# nothing
	}

	public function value_start($parser, $name, $attr)
	{
		$this->value = array('attr' => $attr, 'value' => '');
	}

	public function value_end($parser, $name)
	{
		$this->obj[$this->tag]['list'][] = $this->value;
		unset ($this->value);
	}

	public function value_cdata($parser, $data)
	{
		$this->value['value'] .= $data;
	}

	public function unknown_start($parser, $name, $attr)
	{
		$tags = array();

		foreach ($parser->stack() as $tag) {
			$tags[] = $tag['name'];
		}

		throw new Exception('Unknown start tag: ' . implode('/', $tags));
	}

	public function unknown_end($parser, $name)
	{
		$tags = array();

		foreach ($parser->stack() as $tag) {
			$tags[] = $tag['name'];
		}

		throw new Exception('Unknown end tag: ' . implode('/', $tags));
	}
}

SubversionXMLParser.php

<?php

require_once 'EasyXMLParser.php';

class SubversionXMLParser
{
	public $items;

	function __construct ($filename) {
		$this->items = array();

		$parser = new EasyXMLParser;

		$parser->rule("svn index ?", array(
			'start' => function ($attr, $name) {
				$this->items[$name][] = $attr;
			}
		));

		$parser->parse_file($filename);
	}
}